In an existing search system supporting real-time search, indexing and searching are merged into one process, a new reverse-order sorting index may be used immediately in search. After indexed data in the memory reach a threshold or a time interval reaches a threshold, the indexed data in the memory are merged into indexed data in a disk. Thus in search, both the indexed data in the memory and the indexed data in the disk have to be searched at the same time. Real-time performance of the search system can meet the requirement when a data set to be searched is small. However, as searching and indexing are integrated in one process, both searching and indexing will consume a lot of computational resources. Therefore, in a search system with massive data, adoption of such a searching architecture will consume a lot of hardware equipments, leading to high cost and low efficiency.